Per reporter: Li Menglin Per editor: Lan Suying
"I'm at Computex, I'll make you like me the most, yeah, sing with me, I like NVIDIA the most," NVIDIA CEO Jensen Huang casually composed a few lyrics at an event on May 29, AI completed the composition and singing in seconds, Huang Jensen led thousands of people to sing karaoke together.
For a long time, creative activities have been generally regarded as the most difficult jobs to be replaced by AI, but under the "violent" training of large-scale AI models, human creative activities seem to be decomposed and reconstructed by a large amount of data, and the words "inspiration" and "creativity" have been demystined.
Since ChatGPT completely ignited the fire of generative AI, the evolution of the music field has also accelerated. From lyrics creation and composition to arrangement and singing, AI has permeated all aspects of music production.
A number of industry insiders interviewed by "Daily Economic News" said that AI music has begun to occupy a place in the fields of games and video soundtracks, and some companies can save up to 200,000 costs a year using AI.
When the technical threshold gradually decreased, ordinary people also began to try to make money with music. On station B, Up masters took advantage of the popularity of AI Sun Yanzi to open their brains and train various AI covers; On Youtube, there are bloggers who use AI-generated music with AI illustrations, and the annual advertising revenue can reach up to $100,000.
Midjourney has begun to hit the original artist industry, and will musicians who are also creative workers also fall into anxiety?
In the eyes of industry insiders who actively embrace it, AI will greatly lower the technical threshold of music from idea to finished product, and in the future, individuals can become creators based only on their own musical taste.
Photo source: Photo by Xinhua News Agency reporter Zhang Bowen
Cost savings of up to 200,000
"When doing the metaverse, we need a lot of background music and game soundtracks, such as a window popping up, we need a soundtrack. This kind of soundtrack is not demanding, but it is not good, and AI music is particularly suitable for filling in these gaps. Chen Chenchen, the creator of the Spice SPIICE band and also a metaverse entrepreneur, told the Daily Economic News, "We can't recruit someone specifically to do the soundtrack, because this is a volume thing, and the cost of doing it ourselves is too high." ”
Talking about the impact of AI music technology on the company, Chen Chen Chen said, "(AI) has improved a lot in cost and efficiency, and can recruit many people less. To recruit a person who makes music, you also have to buy a variety of equipment, this set of equipment is estimated to cost 100,000, plus labor (salary), almost 200,000 can be saved. ”
The "OXYZ3 Axis Factory" founded by Chen Chen Chen is committed to creating a metaverse virtual world for musicians, but many colleagues in the company do not actually know how to make music, they only need to select the right AI to generate music and put it in the virtual world.
Looking back at the history of music, recording technology has freed music listening from the shackles of live performance, the emergence of synthesizers and drum machines has replaced live performance, and online downloads have completely destroyed the business model of the traditional recording industry. The rapid evolution of AI technology has once again brought the music industry to what economist Schumpeter called a moment of "creative destruction," the disruptive effects of which are already emerging.
Commercial music does not have high requirements for personalized expression, but it is very sensitive to copyright costs and production costs, so it has become the first batch of scenarios for AI music landing, and a group of companies that provide compliant AI music have also emerged.
"The traditional game soundtrack process is very complicated, usually the planning and art departments first align the requirements, then ask the producer to compose, then arrange, mix, and finally review, the process of constantly modifying, the cost is very high." Yang Jie, head of music products at Chengdu Potential Artificial Intelligence Technology Co., Ltd. (Walker AI), told the "Daily Economic News" reporter, "AI technology can quickly generate demos (music samples) from demand, determine the direction and then refine, eliminating many intermediate links." ”
According to Yang Jie, Walker AI uses deep learning, neural networks and other methods to train music generation models of different styles and different application scenarios, the original intention is to meet the music copyright demands of game customers in the production and marketing process, and later found that a wider range of customers such as advertising, self-media, and even TV series variety shows also have demand.
Ordinary people who rub traffic
New technologies not only bring high benefits at the enterprise level, but also lower the threshold for music creation, allowing ordinary people to participate in it and even monetize.
On Youtube, there is a category of music videos called Lofi (low fidelity) with extremely high traffic because of its simple structure, soothing and relaxing rhythm, and white noise background sound, which is especially suitable for long-term listening during work and study. Lofi Girl, the top blogger in the Lofi music category, can make millions of dollars a year with the advertising share of views alone.
At present, with the help of AI tools, even ordinary people who do not understand music have begun to "make money".
Image source: Youtube screenshot
Youtube channel meloChill only started uploading Lofi music 5 months ago and has amassed more than 2,000 followers. According to the analysis of financial video blogger MrTurner, meloChill's music is generated by AI music software Mubert, and with the image generated by Midjourney as a video cover, a song may be completed in 10 minutes. He calculated that according to the current development trend of meloChill, it can make at least $10,000 a year and a maximum of $100,000.
Image source: Youtube screenshot
During the same period, the B station Up master "Eternity丨L" began uploading videos of AI Sun Yanzi covering songs by Jay Chou, Lin Junjie and others from April 15. According to the statistics of the reporter of the "Daily Economic News", the blogger's 27 such videos have a total of about 7 million views. According to a practitioner of an MCN organization, the current incentive plan of station B can give 0.3 yuan to 0.4 yuan for 1,000 views. Based on this calculation, the blogger earned 2100 yuan to 2800 yuan in more than 1 month by relying on AI Sun Yanzi video.
Where do musicians go in the AI era?
In the face of the explosion of AI Sun Yanzi, Sun Yanzi herself recently issued a response that humans have always believed that ideas or opinions cannot be copied by machines, but AI has begun to threaten industries such as law, medicine and accounting, and although AI synthetic covers are now flawed, they will also be overcome soon.
"Ironically, humanity's inability to surpass it is just around the corner." Sun Yanzi's attitude towards AI reveals a sense of helplessness, and said that "pure thinking, being yourself is enough".
When ChatGPT became popular, some people applied Liu Cixin's concept in "The Three-Body Problem", saying that the attitude towards AI can be roughly divided into the "advent faction" that embraces infinitely, the "rescuer" who holds AI to liberate human productivity, and the "survivor" who doubts and worries. In fact, the attitude of many musicians interviewed by the "Daily Economic News" reporter to AI technology has also been clearly differentiated.
Liu Siqi, a famous music producer with more than 20 years of experience and CEO of Chengdu Lechain Technology, said that he is the one who embraces technology. He believes that ordinary people can now use large models to generate "50% to 60% level" works, without learning "do re mi" or learning to write lyrics, so he is very looking forward to AI music replacing the current explosive "garbage".
"The music made by AI is at least above the horizon. Isn't the music you currently hear on the short video platform more garbage than AI music? Liu Siqi said, "The platform wants traffic, everything must explode, after a popular song comes out, a hundred songs are immediately copied, just change a few tones, not plagiarism." ”
In Liu Siqi's view, only 1% to 2% of the head can make money under the flow logic, and the middle waist and below are all "chicken feathers", and AI music can impact this over-centralized situation.
Chen Chen Chen has both a musician and an entrepreneur, and his attitude towards AI technology is relatively complicated. As early as 2021, the spice band SPIICE collaborated on songwriting with Microsoft's spin-off artificial intelligence business "Xiaoice". Recalling that creative experience, Chen Chen Chen confessed that the work completed from Xiaoice was only the initial "motif" stage of music creation, and could not complete the subsequent work according to the feedback of the musicians.
Despite this, Chen Chen is optimistic about the further development of AI music. "If the AI gives me 30 motivations, I preview and choose one, and the AI expands on that and gives another 30, and so on until a song is finished. I think this makes a lot of sense, because people who can't make music can also make songs through their own tastes. Chen Chen Chen said.
As the technical links in the process of music from the initial "idea" to the final "presentation" are gradually completed by AI tools, the creativity of musicians will be more reflected in inspiration and ideas, and the focus of technology will also become the skillful call of AI tools, so as to achieve "taste" to write songs. This "rescuer" view of AI has partially become a reality.
"Music creators are both data feeders, providing professional music theory knowledge, and model tuners, optimizing algorithms in the continuous training of models." Talking about the creative process of the company's music products, Yang Jie of Walker AI said, "In addition, we use these technologies to help musicians, music producers and music educators improve their work efficiency and work quality, and also bring more possibilities and innovation to the music industry, so that creators can pay more attention to creativity and release creativity." ”
Liu Siqi also believes that in the future music industry, people and AI are a cooperative relationship. "In AI music, the creator bids farewell to the role of a tool man, and makes the AI algorithm an aid for the landing of ideas, and what he really contributes is actually a prompt, and copyright distribution is an incentive for this higher-dimensional creativity and imagination." Liu Siqi said.
Daily economic news